Towards Cross-Granularity Few-Shot Learning: Coarse-to-Fine Pseudo-Labeling with Visual-Semantic Meta-Embedding

TL;DR

This paper introduces a novel cross-granularity few-shot learning approach that uses pseudo-labeling and meta-embedding to enable fine-grained classification with only coarse labels during training, reducing annotation costs.

Contribution

It proposes a coarse-to-fine pseudo-labeling method with a visual-semantic meta-embedder to improve few-shot classification across different label granularities.

Findings

Effective on three datasets

Reduces annotation cost for fine-grained tasks

Robust across various experimental settings

Abstract

Few-shot learning aims at rapidly adapting to novel categories with only a handful of samples at test time, which has been predominantly tackled with the idea of meta-learning. However, meta-learning approaches essentially learn across a variety of few-shot tasks and thus still require large-scale training data with fine-grained supervision to derive a generalized model, thereby involving prohibitive annotation cost. In this paper, we advance the few-shot classification paradigm towards a more challenging scenario, i.e., cross-granularity few-shot classification, where the model observes only coarse labels during training while is expected to perform fine-grained classification during testing. This task largely relieves the annotation cost since fine-grained labeling usually requires strong domain-specific expertise. To bridge the cross-granularity gap, we approximate the fine-grained data distribution by greedy clustering of each coarse-class into pseudo-fine-classes according to the similarity of image embeddings. We then propose a meta-embedder that jointly optimizes the visual- and semantic-discrimination, in both instance-wise and coarse class-wise, to obtain a good feature space for this coarse-to-fine pseudo-labeling process. Extensive experiments and ablation studies are conducted to demonstrate the effectiveness and robustness of our approach on three representative datasets.